The effect of temporal relationship of stimulus compound on ambiguous discrimination in the pigeon's autoshaping.

Pigeons learned a serial amibiguous discrimination task (FA+, A-, FB-, B+), in which the serial presentation of a feature and one target (FA) and the presentation of the other target alone (B) were followed by an unconditioned stimulus (US), but the serial FB presentation and the presentation of A alone were unreinforced. The successful performance of this task supported Holland's (1983, 1985) model of the feature modulation asserting that the feature occasion-sets each target-US association, rather than Rescorla's (1985) model asserting that the feature modulates the threshold of the US representation. The result that the simultaneous presentation of the feature and the target did not produce better discrimination than the serial presentation suggested a 'bidirectional occasion-setting' function of the feature. However, the insertion of the temporal gap between the feature and the target deteriorated the discrimination, a result that disagreed with previous studies of occasion-setting in the rats. Some possible reasons of the detrimental effect of the gap were discussed, and a unique-cue strategy was suggested as a possible alternative account of the discrimination.